distributed particle filter tuning

Existing distributed particle filtering methods are constrained by fixed architectures and rigid communication assumptions, limiting their adaptability in open, heterogeneous Internet of Things (IoT) environments. This work introduces aggregation computing to this domain for the first time, proposing a unified framework based on the computational field abstraction that decouples state estimation from information propagation. This design enables flexible configuration of fusion centers, measurement aggregation schemes, and dissemination strategies. The approach significantly enhances system adaptability and scalability in dynamic IoT settings. Simulation experiments demonstrate effective trade-offs among estimation accuracy, communication overhead, and robustness across various configurations, confirming the framework’s broad applicability to diverse deployment scenarios.

In particle filtering, the “prior boundary phenomenon”—estimation failure when target states exceed the limited support of the prior distribution—severely degrades robustness. Method: This paper proposes Diffusion-Enhanced Particle Filtering (DEPF), a novel framework introducing three core mechanisms: adaptive diffusion-based exploration, entropy-driven weight regularization, and dynamic kernel support expansion—enabling online, controllable adaptation of the prior support set. DEPF integrates diffusion process modeling, information-theoretic entropy constraints, kernel density perturbation, and Bayesian resampling to overcome prior boundary limitations while preserving computational efficiency. Contribution/Results: We provide theoretical convergence guarantees for DEPF. Empirical evaluation demonstrates substantial improvements in estimation success rate and accuracy under high-dimensional and non-convex dynamic scenarios; average estimation error decreases by over 40% compared to state-of-the-art baselines.

This work addresses the challenge of balancing communication efficiency and estimation accuracy in multi-agent cooperative perception by proposing an event-triggered sparsified information diffusion framework (EDC-CIF). The method integrates an error-minimizing event-triggering mechanism with cubature information filtering for local state estimation and employs a correlation-aware diffusion strategy to enable efficient global fusion. Both theoretical analysis and experimental results demonstrate that EDC-CIF overcomes the traditional trade-off between communication overhead and estimation performance, significantly reducing communication volume and computational time while simultaneously improving tracking accuracy and convergence speed. The framework exhibits strong scalability, making it well-suited for large-scale multi-agent systems.

This work addresses the real-time optimization of mobile radar deployment in dynamic environments. Methodologically, it introduces an information-driven trajectory planning framework that uniquely couples Model Predictive Path Integral (MPPI) control with a physically grounded radar range measurement model incorporating actual radar parameters. The framework integrates Cubature Kalman Filter-based state estimation, information-geometric modeling, and Monte Carlo uncertainty quantification to enable kinematically feasible, dynamically consistent, and obstacle-aware online trajectory optimization. Experimental evaluation across 500 simulations demonstrates that the proposed approach reduces average root-mean-square error (RMSE) by 38–74% compared to static deployment and simplified models, while shrinking the upper tail of the 90% highest-density interval by 33–79%. These improvements significantly enhance both localization accuracy and robustness for dynamic targets.

This work proposes a novel particle filtering framework tailored for discrete-time state-space models under severely degraded or vanishing observation noise, specifically when the observation equation is a linear function of the latent state with degenerate additive noise. The method preserves desirable filtering properties even as the noise approaches its degenerate limit and, for the first time, extends this approach to continuous-time models where the hidden state is driven by a diffusion process. By integrating state-space modeling, specialized handling of degenerate noise, and refined time-discretization analysis, the proposed algorithm demonstrates strong robustness and high accuracy in multiple numerical experiments, particularly under low-noise regimes and fine temporal resolutions.

This work addresses the vulnerability of single-vessel tracking from fixed coastal platforms to sensor degradation—specifically, camera performance under adverse lighting and occlusion, and LiDAR limitations at long ranges or with weak reflectivity. To mitigate these issues, the authors propose an adaptive multi-sensor fusion particle filter grounded in information gain (entropy reduction), which dynamically schedules a shore-based 3D LiDAR and an elevated camera to select the optimal sensing configuration within each fusion window. The approach maintains high tracking accuracy while substantially reducing computational overhead by avoiding continuous multi-stream processing. Real-world maritime experiments demonstrate that the system prioritizes LiDAR in near-field scenarios and switches to camera-based tracking in the far field, achieving both continuity and robustness. This provides an efficient and practical tracking baseline for resource-constrained maritime surveillance.

This work addresses Bayesian filtering for continuous-discrete state-space models where the hidden state evolves according to an Itô stochastic differential equation and observations arrive at discrete time instances. The authors propose a novel constrained particle filter that enforces hard support constraints on the state at each observation time via barrier functions, directly restricting the system dynamics rather than truncating the likelihood, thereby enhancing numerical stability. A unified theoretical analysis establishes convergence and time-uniform error bounds that explicitly account for numerical integration errors arising from the SDE solver. Experimental results on the stochastic Lorenz-96 system demonstrate that the proposed method effectively confines the state exploration range while maintaining high accuracy, significantly outperforming conventional particle filters.

This work proposes a novel diffusion Monte Carlo (DMC) implementation, termed DMCD, which replaces the conventional breadth-first traversal and particle ensemble simulation with a depth-first traversal strategy leveraging a stack-based structure. By introducing, for the first time, a stack-oriented approach from particle transport into the DMC framework, DMCD provides a unified treatment of both eigenvalue and linear equation problems. The method naturally accommodates population control and offspring weighting, and effectively addresses the challenge of initializing new walkers upon stack depletion through mechanisms including splitting, Russian roulette, and importance sampling. Experimental results demonstrate that DMCD substantially reduces memory consumption and enhances utilization of memory hierarchies and coprocessors, highlighting its potential as a viable alternative to traditional DMC implementations.